From e3f721e6f6d5d916edf71e5f26ac0547a4b28e1e Mon Sep 17 00:00:00 2001
From: Pavel Begunkov <asml.silence@gmail.com>
Date: Fri, 24 Sep 2021 21:59:48 +0100
Subject: [PATCH] io_uring: split iopoll loop
Git-commit: e3f721e6f6d5d916edf71e5f26ac0547a4b28e1e
Patch-mainline: v5.16-rc1
References: bsc#1205205

The main loop of io_do_iopoll() iterates and does ->iopoll() until it
meets a first completed request, then it continues from that position
and splices requests to pass them through io_iopoll_complete().

Split the loop in two for clearness, iopolling and reaping completed
requests from the list.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/a7f6fd27a94845e5dc925a47a4a9765a92e514fb.1632516769.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Gabriel Krisman Bertazi <krisman@suse.de>
---
 fs/io_uring.c | 32 +++++++++++++++++++-------------
 1 file changed, 19 insertions(+), 13 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 90ad9cd254c5..0e46ed6f92ea 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -2441,6 +2441,12 @@ static void io_iopoll_complete(struct io_ring_ctx *ctx, struct list_head *done)
 	io_req_free_batch_finish(ctx, &rb);
 }
 
+/* same as "continue" but starts from the pos, not next to it */
+#define list_for_each_entry_safe_resume(pos, n, head, member) 		\
+	for (n = list_next_entry(pos, member);				\
+	     !list_entry_is_head(pos, head, member);			\
+	     pos = n, n = list_next_entry(n, member))
+
 static int io_do_iopoll(struct io_ring_ctx *ctx, bool force_nonspin)
 {
 	struct io_kiocb *req, *tmp;
@@ -2456,7 +2462,7 @@ static int io_do_iopoll(struct io_ring_ctx *ctx, bool force_nonspin)
 	if (ctx->poll_multi_queue || force_nonspin)
 		poll_flags |= BLK_POLL_ONESHOT;
 
-	list_for_each_entry_safe(req, tmp, &ctx->iopoll_list, inflight_entry) {
+	list_for_each_entry(req, &ctx->iopoll_list, inflight_entry) {
 		struct kiocb *kiocb = &req->rw.kiocb;
 		int ret;
 
@@ -2465,12 +2471,7 @@ static int io_do_iopoll(struct io_ring_ctx *ctx, bool force_nonspin)
 		 * If we find a request that requires polling, break out
 		 * and complete those lists first, if we have entries there.
 		 */
-		if (READ_ONCE(req->iopoll_completed)) {
-			list_move_tail(&req->inflight_entry, &done);
-			nr_events++;
-			continue;
-		}
-		if (!list_empty(&done))
+		if (READ_ONCE(req->iopoll_completed))
 			break;
 
 		ret = kiocb->ki_filp->f_op->iopoll(kiocb, &iob, poll_flags);
@@ -2481,17 +2482,22 @@ static int io_do_iopoll(struct io_ring_ctx *ctx, bool force_nonspin)
 
 		/* iopoll may have completed current req */
 		if (!rq_list_empty(iob.req_list) ||
-		    READ_ONCE(req->iopoll_completed)) {
-			list_move_tail(&req->inflight_entry, &done);
-			nr_events++;
-		}
+		    READ_ONCE(req->iopoll_completed))
+			break;
 	}
 
 	if (!rq_list_empty(iob.req_list))
 		iob.complete(&iob);
-	if (!list_empty(&done))
-		io_iopoll_complete(ctx, &done);
+	list_for_each_entry_safe_resume(req, tmp, &ctx->iopoll_list,
+					inflight_entry) {
+		if (!READ_ONCE(req->iopoll_completed))
+			break;
+		list_move_tail(&req->inflight_entry, &done);
+		nr_events++;
+	}
 
+	if (nr_events)
+		io_iopoll_complete(ctx, &done);
 	return nr_events;
 }
 
-- 
2.35.3